Functional stress imaging to predict abnormal coronary fractional flow reserve: the PACIFIC 2 study.

AIMS: The diagnostic performance of non-invasive imaging in patients with prior coronary artery disease (CAD) has not been tested in prospective head-to-head comparative studies. The aim of this study was to compare the diagnostic performance of qualitative single-photon emission computed tomography (SPECT), quantitative positron emission tomography (PET), and qualitative magnetic resonance imaging (MRI) in patients with a prior myocardial infarction (MI) or percutaneous coronary intervention (PCI). METHODS AND RESULTS: In this prospective clinical study, all patients with prior MI and/or PCI and new symptoms of ischaemic CAD underwent 99mTc-tetrofosmin SPECT, [15O]H2O PET, and MRI, followed by invasive coronary angiography with fractional flow reserve (FFR) in all coronary arteries. All modalities were interpreted by core laboratories. Haemodynamically significant CAD was defined by at least one coronary artery with an FFR ≤0.80. Among the 189 enrolled patients, 63% had significant CAD. Sensitivity was 67% (95% confidence interval 58-76%) for SPECT, 81% (72-87%) for PET, and 66% (56-75%) for MRI. Specificity was 61% (48-72%) for SPECT, 65% (53-76%) for PET, and 62% (49-74%) for MRI. Sensitivity of PET was higher than SPECT (P = 0.016) and MRI (P = 0.014), whereas specificity did not differ among the modalities. Diagnostic accuracy for PET (75%, 68-81%) did not statistically differ from SPECT (65%, 58-72%, P = 0.03) and MRI (64%, 57-72%, P = 0.052). Using FFR < 0.75 as a reference, accuracies increased to 69% (SPECT), 79% (PET), and 71% (MRI). CONCLUSION: In this prospective head-to-head comparative study, SPECT, PET, and MRI did not show a significantly different accuracy for diagnosing FFR defined significant CAD in patients with prior PCI and/or MI. Overall diagnostic performances, however, were discouraging and the additive value of non-invasive imaging in this high-risk population is questionable.

Sex Differences in the Relationship between New York Heart Association Functional Classification and Survival in Cardiovascular Disease Patients: A Mediation Analysis of Exercise Capacity with Regular Care Data.

Background: The New York Heart Association (NYHA) functional class has extensively been used for risk stratification in patients suspected of heart failure, although its prognostic value differs between sexes and disease entities. Functional exercise capacity might explain the association between NYHA functional class and survival, and can serve as an objective proxy for the subjective nature of the NYHA classification. Therefore, we assessed whether sex-differences in exercise capacity explain the association between NYHA functional class and survival in patients suspected of cardiovascular disease. Methods: Electronic health record data from 7259 patients with cardiovascular symptoms, a documented NYHA functional class and cardiac stress electrocardiogram (ECG), was analysed. Follow-up for all-cause mortality was obtained through linkage with Statistics Netherlands. A sex-stratified mediation analysis was performed to assess to what extent the proportional heart rate and -workload during ECG stress testing explain the association between NYHA functional class and survival. Results: In men, increments in NYHA functional class were related to higher all-cause mortality in a dose-response manner (NYHA II vs III/IV: hazard ratio [HR] 1.59 vs 3.64, referenced to NYHA I), whilst in women those classified as NYHA functional class II and III/IV had a similar higher mortality risk (HR 1.49 vs 1.41). Sex-stratified mediation analysis showed that the association between NYHA and survival was mostly explained by proportional workload during stress ECG (men vs women: 22.9%, 95% CI: 18.9%-27.3% vs 40.3%, 95% CI: 28.5%-68.6%) and less so by proportional heart rate (men vs women: 2.5%, 95% CI: 1.3%-4.3% vs 8.0%, 95% CI: 4.1%-18.1%). Post-hoc analysis showed that NYHA classification explained a minor proportion of the association between proportional workload and all-cause mortality (men vs women: 15.1%, 95% CI: 12.0%-18.3% vs 4.4%, 95% CI: 1.5%-7.4%). Conclusions: This study showed a significant mediation in both sexes on the association between NYHA functional class and all-cause mortality by proportional workload, but the effect explained by NYHA classification on the association between survival and proportional workload is small. This implies that NYHA classification is not a sole representation of a patient's functional capacity, but might also incude other aspects of the patient's overall health status.

Routine clinical care data from thirteen cardiac outpatient clinics: design of the Cardiology Centers of the Netherlands (CCN) database.

BACKGROUND: Despite the increasing availability of clinical data due to the digitalisation of healthcare systems, data often remain inaccessible due to the diversity of data collection systems. In the Netherlands, Cardiology Centers of the Netherlands (CCN) introduced "one-stop shop" diagnostic clinics for patients suspected of cardiac disease by their general practitioner. All CCN clinics use the same data collection system and standardised protocol, creating a large regular care database. This database can be used to describe referral practices, evaluate risk factors for cardiovascular disease (CVD) in important patient subgroups, and develop prediction models for use in daily care. CONSTRUCTION AND CONTENT: The current database contains data on all patients who underwent a cardiac workup in one of the 13 CCN clinics between 2007 and February 2018 (n = 109,151, 51.9% women). Data were pseudonymised and contain information on anthropometrics, cardiac symptoms, risk factors, comorbidities, cardiovascular and family history, standard blood laboratory measurements, transthoracic echocardiography, electrocardiography in rest and during exercise, and medication use. Clinical follow-up is based on medical need and consisted of either a repeat visit at CCN (43.8%) or referral for an external procedure in a hospital (16.5%). Passive follow-up via linkage to national mortality registers is available for 95% of the database. UTILITY AND DISCUSSION: The CCN database provides a strong base for research into historically underrepresented patient groups due to the large number of patients and the lack of in- and exclusion criteria. It also enables the development of artificial intelligence-based decision support tools. Its contemporary nature allows for comparison of daily care with the current guidelines and protocols. Missing data is an inherent limitation, as the cardiologist could deviate from standardised protocols when clinically indicated. CONCLUSION: The CCN database offers the opportunity to conduct research in a unique population referred from the general practitioner to the cardiologist for diagnostic workup. This, in combination with its large size, the representation of historically underrepresented patient groups and contemporary nature makes it a valuable tool for expanding our knowledge of cardiovascular diseases. TRIAL REGISTRATION: Not applicable.

Elevated levels of circulating DNA and chromatin are independently associated with severe coronary atherosclerosis and a prothrombotic state.

OBJECTIVE: Aberrant neutrophil activation occurs during the advanced stages of atherosclerosis. Once primed, neutrophils can undergo apoptosis or release neutrophil extracellular traps. This extracellular DNA exerts potent proinflammatory, prothrombotic, and cytotoxic properties. The goal of this study was to examine the relationships among extracellular DNA formation, coronary atherosclerosis, and the presence of a prothrombotic state. APPROACH AND RESULTS: In a prospective, observational, cross-sectional cohort of 282 individuals with suspected coronary artery disease, we examined the severity, extent, and phenotype of coronary atherosclerosis using coronary computed tomographic angiography. Double-stranded DNA, nucleosomes, citrullinated histone H4, and myeloperoxidase-DNA complexes, considered in vivo markers of cell death and NETosis, respectively, were established. We further measured various plasma markers of coagulation activation and inflammation. Plasma double-stranded DNA, nucleosomes, and myeloperoxidase-DNA complexes were positively associated with thrombin generation and significantly elevated in patients with severe coronary atherosclerosis or extremely calcified coronary arteries. Multinomial regression analysis, adjusted for confounding factors, identified high plasma nucleosome levels as an independent risk factor of severe coronary stenosis (odds ratio, 2.14; 95% confidence interval, 1.26-3.63; P=0.005). Markers of neutrophil extracellular traps, such as myeloperoxidase-DNA complexes, predicted the number of atherosclerotic coronary vessels and the occurrence of major adverse cardiac events. CONCLUSIONS: Our report provides evidence demonstrating that markers of cell death and neutrophil extracellular trap formation are independently associated with coronary artery disease, prothrombotic state, and occurrence of adverse cardiac events. These biomarkers could potentially aid in the prediction of cardiovascular risk in patients with chest discomfort.

High-sensitive troponin T assay for the diagnosis of acute myocardial infarction: an economic evaluation.

BACKGROUND: Delayed diagnosis and treatment of Acute Myocardial Infarction (AMI) has a major adverse impact on prognosis in terms of both morbidity and mortality. Since conventional cardiac Troponin assays have a low sensitivity for diagnosing AMI in the first hours after myocardial necrosis, high-sensitive assays have been developed. The aim of this study was to assess the cost effectiveness of a high-sensitive Troponin T assay (hsTnT), alone or combined with the heart-type fatty acid-binding protein (H-FABP) assay in comparison with the conventional cardiac Troponin (cTnT) assay for the diagnosis of AMI in patients presenting to the hospital with chest pain. METHODS: We performed a cost-utility analysis (quality adjusted life years-QALYs) and a cost effectiveness analysis (life years gained-LYGs) based on a decision analytic model, using a health care perspective in the Dutch context and a life time time-horizon. The robustness of model predictions was explored using one-way and probabilistic sensitivity analyses. RESULTS: For a life time incremental cost of 30.70 Euros, use of hsTnT over conventional cTnT results in gain of 0.006 Life Years and 0.004 QALY. It should be noted here that hsTnT is a diagnostic intervention which costs only 4.39 Euros/test more than the cTnT test. The ICER generated with the use of hsTnT based diagnostic strategy comparing with the use of a cTnT-based strategy, is 4945 Euros per LYG and 7370 Euros per QALY. The hsTnT strategy has the highest probability of being cost effective at thresholds between 8000 and 20000 Euros per QALY. The combination of hsTnT and h-FABP strategy's probability of being cost effective remains lower than hsTnT at all willingness to pay thresholds. CONCLUSION: Our analysis suggests that hsTnT assay is a very cost effective diagnostic tool relative to conventional TnT assay. Combination of hsTnT and H-FABP does not offer any additional economic and health benefit over hsTnT test alone.

Simplification of a registry-based algorithm for ejection fraction prediction in heart failure patients: Applicability in cardiology centres of the Netherlands.

BACKGROUND: Left ventricular ejection fraction (EF) is used to categorize heart failure (HF) into phenotypes but this information is often missing in electronic health records or non-HF registries. METHODS: We tested the applicability of a simplified version of a multivariable algorithm, that was developed on data of the Swedish Heart Failure Registry to predict EF in patients with HF. We used data from 4,868 patients with HF from the Cardiology Centers of the Netherlands database, an organization of 13 cardiac outpatient clinics that operate between the general practitioner and the hospital cardiologist. The algorithm included 17 demographical and clinical variables. We tested model discrimination, model performance and calculated model sensitivity, specificity, positive and negative predictive values for EF ≥ vs. <50% and EF ≥ vs. <40%. We additionally performed a multivariable multinomial analysis for all three separate HF phenotypes (with reduced, mildly reduced and preserved EF) HFrEF vs. HFmrEF vs. HFpEF. Finally, we internally validated the model by using temporal validation. RESULTS: Mean age was 66 ±12 years, 44% of patients were women, 68% had HFpEF, 17% had HFrEF, and 15% had HFmrEF. The C-statistic was of 0.71 for EF ≥/< 50% (95% CI: 0.69-0.72) and of 0.74 (95% CI: 0.73-0.75) for EF ≥/< 40%. The model had the highest sensitivities for EF ≥50% (0.72, 95% CI: 0.63-0.75) and for EF ≥40% (0.70, 95% CI: 0.65-0.71). Similar results were achieved by the multinomial model, but the C-statistics for predicting HFpEF vs HFrEF was lower (0.61, 95% CI 0.58-0.63). The internal validation confirmed good discriminative ability. CONCLUSIONS: A simple algorithm based on routine clinical characteristics can help discern HF phenotypes in non-cardiology datasets and research settings such as research on primary care data, where measurements of EF is often not available.

Identifying plasma proteomic signatures from health to heart failure, across the ejection fraction spectrum.

Circulating proteins may provide insights into the varying biological mechanisms involved in heart failure (HF) with preserved ejection fraction (HFpEF) and reduced ejection fraction (HFrEF). We aimed to identify specific proteomic patterns for HF, by comparing proteomic profiles across the ejection fraction spectrum. We investigated 4210 circulating proteins in 739 patients with normal (Stage A/Healthy) or elevated (Stage B) filling pressures, HFpEF, or ischemic HFrEF (iHFrEF). We found 2122 differentially expressed proteins between iHFrEF-Stage A/Healthy, 1462 between iHFrEF-HFpEF and 52 between HFpEF-Stage A/Healthy. Of these 52 proteins, 50 were also found in iHFrEF vs. Stage A/Healthy, leaving SLITRK6 and NELL2 expressed in lower levels only in HFpEF. Moreover, 108 proteins, linked to regulation of cell fate commitment, differed only between iHFrEF-HFpEF. Proteomics across the HF spectrum reveals overlap in differentially expressed proteins compared to stage A/Healthy. Multiple proteins are unique for distinguishing iHFrEF from HFpEF, supporting the capacity of proteomics to discern between these conditions.

Plasma Proteomic Patterns Show Sex Differences in Early Concentric Left Ventricular Remodeling.

BACKGROUND: Concentric remodeling (cRM) can precede heart failure with preserved ejection fraction (HFpEF), a condition prevalent in women. METHODS: Patients (n=60 593, 54.2% women) visiting outpatient clinics of Cardiology Centers of the Netherlands were analyzed for cRM, HFpEF development, and mortality risk. We studied risk factors for relative wall thickness both sex-stratified and in women and men combined. Biomarker profiling was performed (4534 plasma proteins) in a substudy involving 557 patients (65.4% women) to identify pathways involved in cRM. RESULTS: cRM was present in 23.5% of women and 27.6% of men and associated with developing HFpEF (HR, 2.15 [95% CI, 1.51-2.99]) and mortality risk (HR, 1.09 [95% CI, 1.00-1.19]) in both sexes. Age, heart rate, and hypertension were statistically significantly stronger risk factors for relative wall thickness in women than men. Higher circulating levels of IFNA5 (interferon alpha-5) were associated with higher relative wall thickness in women only. Pathway analysis revealed differential pathway activation by sex and increased expression of inflammatory pathways in women. CONCLUSIONS: cRM is prevalent in approximately 1 in 4 women and men visiting outpatient cardiology clinics and associated with HFpEF development and mortality risk in both sexes. Known risk factors for cRM were more strongly associated in women than men. Proteomic analysis revealed inflammatory pathway activation in women, with a central role for IFNA5. Differential biologic pathway activation by sex in cRM may contribute to the female predominance of HFpEF and holds promise for identification of new therapeutic avenues for prevention and treatment of HFpEF. REGISTRATION: URL: https://www. CLINICALTRIALS: gov; Unique identifier: NCT001747.

The HEARTS partner forum-supporting implementation of HEARTS to treat and control hypertension.

Cardiovascular diseases (CVD), principally ischemic heart disease (IHD) and stroke, are the leading causes of death (18. 6 million deaths annually) and disability (393 million disability-adjusted life-years lost annually), worldwide. High blood pressure is the most important preventable risk factor for CVD and deaths, worldwide (10.8 million deaths annually). In 2016, the World Health Organization (WHO) and the United States Centers for Disease Control (CDC) launched the Global Hearts initiative to support governments in their quest to prevent and control CVD. HEARTS is the core technical package of the initiative and takes a public health approach to treating hypertension and other CVD risk factors at the primary health care level. The HEARTS Partner Forum, led by WHO, brings together the following 11 partner organizations: American Heart Association (AHA), Center for Chronic Disease Control (CCDC), International Society of Hypertension (ISH), International Society of Nephrology (ISN), Pan American Health Organization (PAHO), Resolve to Save Lives (RTSL), US CDC, World Hypertension League (WHL), World Heart Federation (WHF) and World Stroke Organization (WSO). The partners support countries in their implementation of the HEARTS technical package in various ways, including providing technical expertise, catalytic funding, capacity building and evidence generation and dissemination. HEARTS has demonstrated the feasibility and acceptability of a public health approach, with more than seven million people already on treatment for hypertension using a simple, algorithmic HEARTS approach. Additionally, HEARTS has demonstrated the feasibility of using hypertension as a pathfinder to universal health coverage and should be a key intervention of all basic benefit packages. The partner forum continues to find ways to expand support and reinvigorate enthusiasm and attention on preventing CVD. Proposed future HEARTS Partner Forum activities are related to more concrete information sharing between partners and among countries, expanded areas of partner synergy, support for implementation, capacity building, and advocacy with country ministries of health, professional societies, academy and civil societies organizations. Advancing toward the shared goals of the HEARTS partners will require a more formal, structured approach to the forum and include goals, targets and published reports. In this way, the HEARTS Partner Forum will mirror successful global partnerships on communicable diseases and assist countries in reducing CVD mortality and achieving global sustainable development goals (SDGs).

Patients using vitamin K antagonists show increased levels of coronary calcification: an observational study in low-risk atrial fibrillation patients.

AIMS: Vitamin K antagonists (VKA) are currently the most frequently used drug to prevent ischaemic stroke in atrial fibrillation (AF) patients. However, VKA use has been associated with increased vascular calcification. The aim of this study was to investigate the contribution of VKA use to coronary artery calcification in low-risk AF patients. METHODS AND RESULTS: A prospective coronary calcium scan was performed in 157 AF patients without significant cardiovascular disease (108 males; mean age 57 ± 9 years). A total of 71 (45%) patients were chronic VKA users. The duration of VKA treatment varied between 6 and 143 months (mean 46 months). No significant differences in clinical characteristics were found between patients on VKA treatment and non-anticoagulated patients. However, median coronary artery calcium scores differed significantly between patients without and patients with VKA treatment [0, inter-quartile range (IQR) 0-40, vs. 29, IQR 0-184; P = 0.001]. Mean coronary calcium scores increased with the duration of VKA use (no VKA: 53 ± 115, 6-60 months on VKA: 90 ± 167, and >60 months on VKA: 236 ± 278; P < 0.001). Multivariable logistic regression analysis revealed that age and VKA treatment were significantly related to increased coronary calcium score. CONCLUSION: Patients using VKA show increased levels of coronary calcification. Age and VKA treatment were independently related to increased coronary calcium score.

[Coronary calcium CT-scanning advisable in patients with a high cardiovascular risk: refining cardiovascular risk in primary prevention]. / Wanneer is de coronaire calciumscore zinvol?

Currently, risk prediction models like SCORE are used for decision making in the primary prevention of cardiovascular disease. The external validity of these models is questionable since they give rise to overtreatment with statins or antihypertensive drugs. Detailed individual risk assessment may reduce this drawback and will increase cost effectiveness. The CT derived coronary calcium score, in asymptomatic patients, was shown to be more accurate than the current prediction models. A coronary calcium score of zero reclassifies a significant number of individuals to a lower risk group and subsequently prevent overtreatment. Using this strategy, it can be anticipated that the Dutch healthcare costs can be reduced by at least 15 million Euro per year.

Statins are associated with a large reduction in all-cause mortality in women from a cardiac outpatient population.

OBJECTIVES: Uncertainty about the benefit of (high-intensity) statins for women remains due to under-representation of women in primary prevention trials and scarcity of sex-stratified data. This study evaluates the sex-specific relation between statin treatment and survival and the additional benefit of high-intensity statins. METHODS: Electronic health record data from 47 801 patients (17 008 statin users and 30 793 non-users) without prior cardiovascular disease were extracted from thirteen Dutch outpatient cardiology clinics. Patients prescribed statins at baseline were propensity-score matched to those eligible for statin therapy (low-density lipoprotein >2.5 mmol/L) without a statin prescription. Statins were divided into low-intensity and high-intensity according to Dutch guidelines. Mortality data were obtained via linkage to the national mortality registry. Cox regression was used to evaluate the relationship between statin prescription and intensity and all-cause and cardiovascular mortality. RESULTS: Propensity score matching created a cohort of 8631 statin users and 8631 non-users. 35% of women and 28% of men received a low-intensity statin. The beneficial effect of statins on both all-cause and cardiovascular mortality was stronger in women (HR 0.66, 95% CI 0.58 to 0.74 and HR 0.55, 95% CI 0.39 to 0.71, respectively) than in men (HR 0.89, 95% CI 0.81 to 0.95 and HR 0.93, 95% CI 0.77 to 1.08, respectively). High-intensity statins conferred modest protection against all-cause mortality (HR 0.94, 95% CI 0.88 to 1.00) and cardiovascular mortality (HR 0.86, 95% CI 0.74 to 0.98) in both sexes. CONCLUSIONS: The protective effect of primary prevention statins was stronger in women than men for both all-cause and cardiovascular mortality. High-intensity statins conferred a modest additional benefit in both sexes. Statins seem to be effective regardless of treatment intensity, especially in women.

Development of a Pipeline for Adverse Drug Reaction Identification in Clinical Notes: Word Embedding Models and String Matching.

BACKGROUND: Knowledge about adverse drug reactions (ADRs) in the population is limited because of underreporting, which hampers surveillance and assessment of drug safety. Therefore, gathering accurate information that can be retrieved from clinical notes about the incidence of ADRs is of great relevance. However, manual labeling of these notes is time-consuming, and automatization can improve the use of free-text clinical notes for the identification of ADRs. Furthermore, tools for language processing in languages other than English are not widely available. OBJECTIVE: The aim of this study is to design and evaluate a method for automatic extraction of medication and Adverse Drug Reaction Identification in Clinical Notes (ADRIN). METHODS: Dutch free-text clinical notes (N=277,398) and medication registrations (N=499,435) from the Cardiology Centers of the Netherlands database were used. All clinical notes were used to develop word embedding models. Vector representations of word embedding models and string matching with a medical dictionary (Medical Dictionary for Regulatory Activities [MedDRA]) were used for identification of ADRs and medication in a test set of clinical notes that were manually labeled. Several settings, including search area and punctuation, could be adjusted in the prototype to evaluate the optimal version of the prototype. RESULTS: The ADRIN method was evaluated using a test set of 988 clinical notes written on the stop date of a drug. Multiple versions of the prototype were evaluated for a variety of tasks. Binary classification of ADR presence achieved the highest accuracy of 0.84. Reduced search area and inclusion of punctuation improved performance, whereas incorporation of the MedDRA did not improve the performance of the pipeline. CONCLUSIONS: The ADRIN method and prototype are effective in recognizing ADRs in Dutch clinical notes from cardiac diagnostic screening centers. Surprisingly, incorporation of the MedDRA did not result in improved identification on top of word embedding models. The implementation of the ADRIN tool may help increase the identification of ADRs, resulting in better care and saving substantial health care costs.

Outcomes in patients with a first episode of chest pain undergoing early coronary CT imaging.

OBJECTIVES: To investigate the impact of a CT-first strategy on all-cause and cardiovascular mortality in patients presenting with chest pain in outpatient cardiology clinics. METHODS: Patients with a first presentation of suspected angina pectoris were identified and their data linked to the registrations of Statistics Netherlands for information on mortality. The linked database consisted of 33 068 patients. CT-first patients were defined as patients with a CT calcium score and coronary CT angiography, within 6 weeks after their initial visit. Propensity score matching (1:5) was used to match patients with and without a CT-first strategy. After matching, 12 545 patients were included of which 2308 CT-first patients and 10 237 patients that underwent usual care. RESULTS: Mean age was 57 years, 56.3% were women and median follow-up was 4.9 years. All-cause mortality was significantly lower in CT-first patients (n=43, 1.9%) compared with patients without CT (n=363, 3.5%) (HR: 0.51, 95% CI 0.37 to 0.70). Furthermore, CT-first patients were more likely to receive cardiovascular preventative and antianginal medication (aspirin: 44.9% vs 27.1%, statins: 48.7% vs 30.3%, beta-blockers: 37.8% vs 25.5%, in CT-first and without CT-first patients, respectively) and to undergo downstream diagnostics and interventions (coronary interventions: 8.5% vs 5.7%, coronary angiography: 16.2% vs 10.6% in CT-first and without CT-first patients, respectively). CONCLUSIONS: In a real-world regular care database, a CT-first strategy in patients suspected of angina pectoris was associated with a lowering of all-cause mortality.

Coronary calcification measures predict mortality in symptomatic women and men.

OBJECTIVE: To assess the prognostic value of absolute and sex-specific, age-specific and race/ethnicity-specific (Multi-Ethnic Study of Atherosclerosis, MESA) percentiles of coronary artery calcification in symptomatic women and men. METHODS: The study population consisted of 4985 symptomatic patients (2793 women, 56%) visiting a diagnostic outpatient cardiology clinic between 2009 and 2018 who were referred for cardiac CT to determine Coronary Artery Calcium Score (CACS). Regular care data were used and these data were linked to the databases of Statistics Netherlands for all-cause mortality data. Kaplan-Meier curves, multivariate Cox proportional hazards regression and concordance statistics were used to evaluate the prognostic value of CACS and MESA percentiles. Women were older compared with men (60 vs 59 years). RESULTS: Median CACS was 0 (IQR: 0-54) in women and 42 (IQR: 0-54) in men. After a median follow-up of 4.4 years (IQR: 3.1-6.3), 116 (2.3%; 53 women and 63 men) patients died. MESA percentiles did not perform better compared with absolute CACS (C-statistic 0.65, 95% CI 0.57 to 0.73, vs 0.66, 95% CI 0.58 to 0.74, in women and 0.59, 95% CI 0.51 to 0.67, vs 0.62, 95% CI 0.55 to 0.69, in men, for the percentiles and absolute CACS, respectively). CONCLUSIONS: In symptomatic individuals absolute CACS predicts mortality with a moderately good performance. MESA percentiles did not perform better compared with absolute CACS, thus there is no need to use them. Including degree of stenosis in the model might slightly improve mortality risk prediction in women, but not in men.

Deep neural networks reveal novel sex-specific electrocardiographic features relevant for mortality risk.

Aims: Incorporation of sex in study design can lead to discoveries in medical research. Deep neural networks (DNNs) accurately predict sex based on the electrocardiogram (ECG) and we hypothesized that misclassification of sex is an important predictor for mortality. Therefore, we first developed and validated a DNN that classified sex based on the ECG and investigated the outcome. Second, we studied ECG drivers of DNN-classified sex and mortality. Methods and results: A DNN was trained to classify sex based on 131 673 normal ECGs. The algorithm was validated on internal (68 500 ECGs) and external data sets (3303 and 4457 ECGs). The survival of sex (mis)classified groups was investigated using time-to-event analysis and sex-stratified mediation analysis of ECG features. The DNN successfully distinguished female from male ECGs {internal validation: area under the curve (AUC) 0.96 [95% confidence interval (CI): 0.96, 0.97]; external validations: AUC 0.89 (95% CI: 0.88, 0.90), 0.94 (95% CI: 0.93, 0.94)}. Sex-misclassified individuals (11%) had a 1.4 times higher mortality risk compared with correctly classified peers. The ventricular rate was the strongest mediating ECG variable (41%, 95% CI: 31%, 56%) in males, while the maximum amplitude of the ST segment was strongest in females (18%, 95% CI: 11%, 39%). Short QRS duration was associated with higher mortality risk. Conclusion: Deep neural networks accurately classify sex based on ECGs. While the proportion of ECG-based sex misclassifications is low, it is an interesting biomarker. Investigation of the causal pathway between misclassification and mortality uncovered new ECG features that might be associated with mortality. Increased emphasis on sex as a biological variable in artificial intelligence is warranted.